The proposal for a satellite mission to observe the height of clouds won first prize at the UNISEC Gobal conference in Japan in November 2014. The proposed constellations of small satellites uses steroscopic observation for the creation of 3D of volcano and other clouds. The idea has convinced the prestigious international commission due to the high usability and developed technical solutions.
E.02 International awards
The PhD thesis has been partialy performed in the frame of the project, its author is a member of the project group, the project leader is the thesis co-supervisor. In the dissertation we have developed a fully automated procedure for orthorectification of optical satellite images. The resulting orthorectified image is in the national coordinate system and constitutes a suitable source for spatial analyses. The original procedure connects several different methods into a single, robust system for automatic generation of orthoimages. The doctoral dissertation describes the whole automatic orthorectification process, which comprises four basic modules: module for extracting and preparing the metadata, module for automatic extraction of ground control points, module for calculation of parameters of the geometric model, and module for orthorectification. Experiments and results with RapidEye and WorldView-2 images are also presented. The experiments evaluate the procedure for automatic extraction of points, the geometric model, the elimination of gross errors, and the positional accuracy of the orthoimages. The results indicate that the automated procedure produces orthoimages with a positional accuracy of about two pixels or better, even if several gross errors are present among the automatically extracted ground control points.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 7257697The proposed invention refers to a method of determination of coordinates of at least one ground control point on the remotely sensed data, such as satellite imagery, by comparing the coordinates of the second point, for example the road junction, on the reference source. By the process according to the invention, first the obtained satellite imagery is rasterized into several layers, after which at least one working layer is selected. Thereafter, objects on the working layer are filtered to obtain the information of digital layer in binary form. These filtered objects are stored into binary file to be added to the reference source as addtional layer. The reference source is converted from vector shapefile to a raster distance image. This is followed by the approximate placement of the digital image in the binary shape on a raster reference source and the exast placement of the image in the binary form on the raster reference source, resulting in a list of ground control points coordinates.
F.33 Slovenian patent
COBISS.SI-ID: 38047237Krištof Oštir was the leader of the international project Sentinel2Agri4Slovenia - Application of Sentinel-2 Time-Series Data for Crop Identification and Crop Stress Monitoring between 2015 and 2017, which was run by Centre of Excellence SPACE-SI, Sinergise and Faculty of Computer and Information Science (University of Ljubljana) as project partners in the framework of PECS program, the European Space Agency (ESA) program for the participating countries. In the project, the research team studies the Sentinel-2 satellite data time-series for crop mask determination, crop identification, and crop stress detection. The final report covers the description of the various high resolution satellite data used (RapidEye, SPOT5, and Sentinel-2), as well as reference and ancillary data from a range of sources and their integration. The report continues with a description of six major crop types (wheat, maize, rapeseed, barley, triticale, and oil pumpkin) investigated in the project with a special focus on the understanding of contribution and correlation of a variety of vegetation indices for crop type separability in a growing season. Five mapping unit types were examined: pixel, pre-defined grid, segment, graphical unit of agriculture land (GERK), and an intersection of GERK and segment. We found the GERK to be the optimal mapping unit. In order to show the results and their potential practical use, a dissemination tool was developed along the implementation of different classification approaches (classical supervised classification and statistical modelling). The report states that the overall accuracy of classification based of Sentinel-2 2016 time-series is 97% for crop mask and 89% for crop type.
D.06 Final report on a foreign/international project
The accessibility of high spatial and high temporal satellite imagery time series has enabled the development of methods of multitemporal land cover classification, which should improve the quality of classification due to time information. The dissertation examines two multitemporal classifications: quasi-multitemporal classification and time series based classification. In the first one, image time series are used as the attributes of single-date classification, while the second one compares the development of certain spectral characteristics over time. The classifications of five basic land cover classes (forest, grass, arable land, water, built-up) and six basic crops (maize, wheat, barley, pumpkin, rapeseed, triticale) are performed using various input images, attributes, mapping units and images of different sensors. Emphasis is given to segmentation processes since the segments as basic units of multitemporal classification are not well researched globally. In addition, several possible effects on the classification result are analysed in detail to provide guidelines for obtaining high precision in short processing time. The results show that satellite image acquisition time, besides spectral values, is the most important attribute in the classification. The quasi-multitemporal classification returns a much higher overall accuracy (+8% basic classes, +16% crops), with an average total accuracy of 90% (basic classes) and 88% (crops) giving the useful operational value. The results of the time series based classification are worse (-1% basic classes, -25% crops), the processing time being extremely long, which makes the method as currently unacceptable for practical use. An important finding of this dissertation is that segments are not the most suitable mapping units for the multitemporal classification. Regardless of the production process, segments give worse classification accuracy than reference polygons (-5% basic classes, -18% crops) and pixels (-5% basic classes, -16% crops).
D.09 Tutoring for postgraduate students